BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:America/Chicago BEGIN:DAYLIGHT DTSTART:20260308T030000 TZOFFSETFROM:-0600 TZOFFSETTO:-0500 RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3 TZNAME:CDT END:DAYLIGHT BEGIN:STANDARD DTSTART:20261101T010000 TZOFFSETFROM:-0500 TZOFFSETTO:-0600 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11 TZNAME:CST END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20260822T012257Z UID:FCA04474-03F1-4246-B0BE-F846A7788254 DTSTART;TZID=America/Chicago:20260820T160000 DTEND;TZID=America/Chicago:20260820T170000 DESCRIPTION:Abstract: The rapid emergence of generative AI and vision-langu age models (VLMs) is transforming multimodal perception and reasoning\, bu t deployment in real-world edge systems remains constrained by bandwidth\, memory movement\, and energy. I will present a near-sensor software–har dware co-design framework that moves early feature extraction and compress ion closer to where data are acquired. Instead of treating the sensor\, me mory\, and processor as separate stages\, we co-optimize models and hardwa re to form compact\, task-specific representations at the sensor boundary. This reduces off-chip transfer and DRAM energy while preserving downstrea m accuracy\, enabling more scalable edge intelligence for vision and multi modal AI. I will then show how this broader principle extends beyond visua l sensing to AI for healthcare. In wearable health AI\, efficient inferenc e requires models that can adaptively process multimodal physiological sig nals\, activate sensing only when needed\, and remain reliable under incom plete observations. Looking forward\, this points toward closed-loop medic al systems that not only sense and infer efficiently\, but also guide time ly actions or interventions. Finally\, I will discuss trustworthy clinical AI\, where generative models for medical reports must be evaluated not on ly for fluency\, but also for consistency\, factual grounding\, and verifi able reasoning. Together\, these directions outline a path toward efficien t\, adaptive\, and trustworthy AI systems across edge and healthcare appli cations.\n\nCo-sponsored by: Rice University ECE and Rice IEEE HKN and Stu dent Branch\n\nSpeaker(s): Gourav Datta\n\nAgenda: \nPresentation at 4:00 to 5:00 pm CDT at Duncan Hall\, Room 3076 at Rice University\n\nRoom: 3076 \, Bldg: Duncan Hall\, Rice University\, 6100 Main Street\, Houston\, Texa s\, United States\, 77005 LOCATION:Room: 3076\, Bldg: Duncan Hall\, Rice University\, 6100 Main Stree t\, Houston\, Texas\, United States\, 77005 ORGANIZER:cavallar@rice.edu SEQUENCE:62 SUMMARY:Near-Sensor Computing and Model–Hardware Co-Design for Energy-Eff icient Vision\, Generative AI\, and Healthcare URL;VALUE=URI:https://events.vtools.ieee.org/m/573368 X-ALT-DESC:Description: <br /><p><strong>Abstract</strong>: The rapid emerg ence of generative AI and vision-language models (VLMs) is transforming mu ltimodal perception and reasoning\, but deployment in real-world edge syst ems remains constrained by bandwidth\, memory movement\, and energy. I wil l present a near-sensor software&ndash\;hardware co-design framework that moves early feature extraction and compression closer to where data are ac quired. Instead of treating the sensor\, memory\, and processor as separat e stages\, we co-optimize models and hardware to form compact\, task-speci fic representations at the sensor boundary. This reduces off-chip transfer and DRAM energy while preserving downstream accuracy\, enabling more scal able edge intelligence for vision and multimodal AI. I will then show how this broader principle extends beyond visual sensing to AI for healthcare. In wearable health AI\, efficient inference requires models that can adap tively process multimodal physiological signals\, activate sensing only wh en needed\, and remain reliable under incomplete observations. Looking for ward\, this points toward closed-loop medical systems that not only sense and infer efficiently\, but also guide timely actions or interventions. Fi nally\, I will discuss trustworthy clinical AI\, where generative models f or medical reports must be evaluated not only for fluency\, but also for c onsistency\, factual grounding\, and verifiable reasoning. Together\, thes e directions outline a path toward efficient\, adaptive\, and trustworthy AI systems across edge and healthcare applications.</p>\n<p><span style="f ont-size: 11.0pt\; font-family: 'Aptos'\,sans-serif\; mso-fareast-font-fam ily: Aptos\; mso-fareast-theme-font: minor-latin\; mso-bidi-font-family: A ptos\; mso-ansi-language: EN-US\; mso-fareast-language: EN-US\; mso-bidi-l anguage: AR-SA\;"><br style="mso-special-character: line-break\;"><!-- [if !supportLineBreakNewLine]--><br style="mso-special-character: line-break\ ;"><!--[endif]--></span></p><br /><br />Agenda: <br /><p>Presentation at 4 :00 to 5:00 <span class="lw_end_time">pm</span>&nbsp\;<span class="lw_cal_ tz_abbrv tz_editable">CDT at Duncan Hall\, Room 3076 at Rice University</s pan></p> END:VEVENT END:VCALENDAR